Pressure ulcers are an important issue for U.S. hospitals. In 2008, the Centers for Medicare and Medicaid Services (CMS) enacted nonpayment policy for hospital stays associated with pressure ulcers not present on admission. This policy produced strong incentives for hospitals to detect pressure ulcers that are present on admission and to implement prevention protocols to reduce pressure ulcers. Evidence-based protocols typically begin with patient risk-assessment shortly following admission using an instrument such as the Braden Scale, with follow-up reassessment of pressure ulcer risk using the Braden Scale every 1-2 days. However, follow-up reassessment with the Braden Scale is costly. Moreover, it is often poorly adhered to; perhaps because clinicians believe there is little value i reassessment and have competing responsibilities for patient care. This suggests that the cost of, and adherence to, Braden Scale assessment might be improved by a dynamic quality improvement strategy that better predicts changes in patient pressure ulcer risk. Such predictors of Braden scores might be based on electronic health record (EHR) data containing prior Braden scores and other variables. The objective of this study is to use EHR data to develop models that can dynamically predict when reassessment of patient Braden scores is likely to lead to a change in pressure ulcer risk in order to develop more effective, and cost-effective, strategies to target Braden Scale risk-assessment for pressure ulcer prevention protocols. Clinicians could use these predictors and past Braden scores to stratify patients for follow-up reassessment. Predictors of Braden scores will be obtained using EHR data from the University of Chicago, which reflects high-quality data from a nursing program that has received ANCC Magnet recognition and is very generalizable to the U.S. The study has two specific aims. The first specific aim is to use EHR data, including comorbid conditions, laboratory data, pharmacy prescriptions and prior Braden scores to develop a model to dynamically predict changes in Braden scores that place patients at high-risk of developing pressure ulcers. This predictive model can be constructed as a mixed-effects regression model using longitudinal data. Multiple predictive models can be constructed from available EHR data and then judged for accuracy in predicting changes in Braden scores. The second specific aim is to develop a decision model to predict the costs, effectiveness, and cost- effectiveness of dynamic strategies to predict changes in Braden scores that could improve the efficiency of and increase the adherence to prescribed Braden Scale reassessment. The model will be built following a focused clinical experience. Included in this analysis will be estimates of how a dynamic strategy might affect adherence to Braden Scale reassessment for patients predicted as high-risk for developing pressure ulcers. It is hoped that greater sensitivity and specificity in predicting risk would redue the costs and improve the effectiveness of pressure ulcer prevention for hospitalized patients.